ing to the different groups in Mendelejeff's table are seen to occupy the same relative positions upon the different portions of this curve. Thus in Group I. the elements Li, Na, K, Rb, Cs, are all found upon the maxima of the curve, and Cu, Ag, and Au at those points at the minima where the electro-negative properties reappear. The halogen elements (chlorine, bromine, iodine) are seen in similar positions upon the ascending, and the alkaline earths (beryllium, magnesium, calcium, strontium, barium) on the descending portions.

When the periodic law was first formulated by Mendelejeff (1869), there were a number of instances in which the system did not harmonise with the then accepted atomic weights of the elements. The discoverer boldly asserted that the atomic weights, and not the system, were at fault, and in almost every such case the careful reinvestigation of the atomic weights by numerous chemists has proved the correctness of the assertion. One or two instances may be quoted. The element indium had assigned to it the atomic weight 76. Its combining proportion is 38, and being regarded as a divalent element, its oxide was believed to have the formula InO. Having an atomic weight = 76, indium would occupy a place between As = 75 and Se 79; but in the system (see table on page 117) there is no room for an element with such an atomic weight; and, moreover, if indium be a divalent element having this atomic weight, it should come between Zn

=

=

65 and

Sr = 87 in Group II., where again there is no room. Mendelejeff made the assumption that the oxide of indium had the formula In2O3, believing the element to be an analogue of aluminium (Group III.). If this be the true composition of the oxide, the atomic weight of the element would be 38 × 3 = 114, and indium would then take its place in Group III., between the elements cadmium 112 and Sn = 118, in the odd series of the second long period. Bunsen afterwards determined the specific heat of indium by means of his ice calorimeter, and found it to be 0.057 :

=

Mean atomic heat 6.4 = 112.3= = atomic weight (see page 47). Specific heat

0.057

Hence 114 and not 76 is the accepted (approximate) atomic weight of indium.

Again, the element beryllium (formerly known as glucinum) has a combining proportion of 4.6. Its chloride was believed to have the composition BeCl, and its oxide to be a sesquioxide having

the formula Be2O3. The atomic weight assigned to the element, therefore, was 13.8.

=

With this atomic weight beryllium would take its place between carbon = 12 and nitrogen 14; but according to the periodic classification there is no room for such an element, and, moreover, in such a position it would be among elements with which it has no properties in common. On the supposition that the oxide of beryllium has the formula BeO, that is, that the element is divalent, its atomic weight would have to be lowered from 13.8 to 9.1 in order to maintain the same ratio between the weights of metal and oxygen in the compound. On this assumption, beryllium would fall into the second place in the first series, between lithium = 7 and boron = 11, and in the same group as magnesium and zinc.

When the specific heat of beryllium was determined, it gave the value 0.45, and this number divided into the atomic heat constant, 6.4, gave 14 as the atomic weight. In spite of this evidence in favour of the higher value as the atomic weight of beryllium, Mendelejeff still regarded the lower number as correct, and it was suggested that possibly beryllium, like carbon and boron (elements also of very low atomic weight), had an abnormally low specific heat at ordinary temperatures. This was found to be the case (see page 48), and at 500° the specific heat of beryllium was found to be 0.6206. This divided into 6.4 gives the value 10 as the atomic weight, which indicates that 9.1 and not 13.8 is in reality the atomic weight of beryllium.

Not only has the periodic law been of service in bringing about the correction of a number of doubtful atomic weights, but by means of it its originator was enabled to predict with considerable certainty the existence of hitherto undiscovered elements, and even to predicate many of the properties of these elements. As already mentioned, at the time when the periodic law was first formulated, there were three gaps in the system in the first long period, namely, No. 4 in the even series (now occupied by scandium), and Nos. 3 and 4 in the odd series (now filled by gallium and germanium). To the unknown elements which were destined to occupy these positions, Mendelejeff gave the names eka-boron, eka-aluminium, and eka-silicon (the prefix eka being the Sanscrit numeral one), and from the known properties of the neighbouring elements of the series (horizontal rows in the table, page 117), and also of those situated nearest in the same family (vertical columns), he predicted some of the prominent properties that would pro

bably be possessed by these elements. Thus in the case of ekaaluminium, from the known properties of aluminium and indium, the neighbouring elements in the same family, and from zinc, the contiguous element in the same series (the 4th place in the series being unoccupied), Mendelejeff deduced the following properties for the unknown element that he called eka-aluminium :—

PREDICTED PROPERTIES OF EKA-ALUMINIUM (1871). (1.) Should have an atomic weight about 69. (2.) Will have a low melting-point.

(3.) Its specific gravity should be about 5.9. (4.) Will not be acted upon by the air.

(5.) Will decompose water at a red heat.

(6.) Will give an oxide ElO,, a chloride ElCl, and sulphate El2(SO4)3

(7.) Will form a potassium alum, which will probably be more soluble and less easily crystallisable than the corresponding aluminium alum.

(8.) The oxide should be more easily reducible to the metal than alumina. The metal will probably be more volatile than aluminium, and therefore its discovery by means of the spectroscope may be expected.

In the year 1875 Lecoq de Boisbaudran discovered a new element in a certain specimen of zinc blende (zinc sulphide), the individuality of which he first recognised by the spectroscope, the spectrum being characterised by a brilliant violet line. This element he named gallium. The properties of this metal, as they were subsequently observed, showed that it was, in fact, the predicted eka-aluminium of Mendelejeff, as will at once be seen by a comparison of the following facts.

PROPERTIES OF GALLIUM (discovered 1875).

(1.) Atomic weight=69.9.

(2.) Melting-point, 30.15°.

(3) Specific gravity, 5.93.

(4.) Only slightly oxidised at a red heat.

(5.) Decomposes water at high temperatures.

(6.) Gallium oxide, GaO3. Gallium chloride, GaCl. Gallium

sulphate, Ga(SO4)3.

(7.) Forms a well-defined alum.

(8.) Is easily obtained by the electrolysis of alkaline solutions.

In a similar manner the properties of eka-boron and eka-silicon were predicted, and the subsequent discovery of scandium (Nilson, 1879), and germanium (Winkler, 1886), whose properties were found to closely accord with these hypothetical elements, formed an additional demonstration of the truth of the periodic law.

There are at present two elements, however, which appear not to conform strictly to this periodic classification. These are the elements argon and tellurium. The atomic weight of argon according to most recent determination is 39.92, while that of potassium is 39.15. Now the periodic system requires that the atomic weight of argon shall be below and not above that of potassium. Again, the latest determinations of the atomic weight of tellurium give 127.6, as against 126.85 for iodine; while in order to conform to the periodic system the atomic weight of tellurium should be below that of iodine. Whether these two cases will prove to be true exceptions, or whether future investigations will show that the atomic weights here given are not the true ones, time alone will show. It must be borne in mind, however, that both argon and tellurium are elements which it is extremely difficult to obtain in a state of absolute purity, and there is considerable probability that in the latter case the element in a pure state has never yet been obtained.

The position which should be given to hydrogen in the periodic system has been the subject of much discussion. It will be noticed that in the table it is placed with a query in Group I. and again in Group VII. ; its univalent character suiting either position equally well. The chief argument in favour of placing it in Group I. is its electro-positive character, in which it strongly resembles the elements lithium, sodium, potassium, &c., metals which may be substituted for hydrogen atom for atom; the "salts of hydrogen" (i.e. acids), and the metallic salts resembling each other when regarded from the ionic standpoint.

The arguments in favour of assigning it a position at the head of Group VII. are more numerous, and may be briefly summarised as follows:-* *

1. Its gaseous character and low boiling-point.

2. Absence of any metallic properties.

3. The diatomic nature of its molecules H2 (while many of the alkali metals are monatomic).

* Masson, Chem. News, vol. lxxiii., p. 283.